Modeling contaminant transport in soil with a transfer function method
Abstract
A transfer function approach to modeling of contaminant transport in soils was developed. A second order linear model was proposed as a means of relating observed experimental output from transport studies to three model parameters which described the soil/contaminant system. Similarities between the Laplace transform and the Moment Generating Function (MGF) allowed development of unique relationships between the three parameters and the first three moments about the origin for experimental output curves. These relationships applied under specific conditions, including: contaminant input to the system was an approximation of an impulse function, and computation of the parameters resulted in an overdamped system. Laboratory column experiments were conducted to provide calibration and verification data for the model. Experiments were conducted with the herbicide alachlor and a composite soil under saturated flow conditions. Several tests were conducted with a single short pulse of alachlor applied to a variety of soil columns, including: three column lengths, four initial moisture contents, and three application concentrations. Two column experiments were conducted with a series of alachlor pulses applied. In all experiments concentrations of the pesticide in the output were analyzed using High Performance Liquid Chromatography and the output distributions determined over time. Model parameters for each of the single pulse input experiments were determined using the moments of the output distribution and the relationships developed. These parameters were then used to compare predicted output with observed output, with reasonable results. Parameters were examined for physical interpretation and changes in parameters with changes in experimental conditions. Parameters developed from single input columns were used in predicting output from the columns with complex inputs with good results. Data taken from the literature were used in application of the model to non-impulse input experimental data. A differentiation technique was used to develop transfer functions for this additional data. Comparison of predicted output and observed output showed good correspondence. Attempts at series combination of short columns showed that attenuation with length was not a linear process.
Degree
Ph.D.
Advisors
Monke, Purdue University.
Subject Area
Agricultural engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.